AI, artificial intelligence, regulation and legislation are constantly evolving to match the pace of innovation. The following provides a brief summary and links to statements from regulatory bodies and pending legislation as it attempts to keep up with the pace of change in this space. Already we see conflicts between regulators and legislators struggling to shape their own visions of proper use of AI. This blog post will be updated frequently to stay current with artificial intelligence developments. Check back frequently to stay on top of the AI landscape.
United States
AI BILL OF RIGHTS
The Whitehouse has offered one possible vision for an AI bill of rights focusing on transparency and the ability to opt out of unwanted uses of AI as it relates to personal data. https://www.whitehouse.gov/ostp/ai-bill-of-rights/
AI EVALUATION
S.4178 - Introduced on April 18, 2024. The text of this bill is still not available, but the bill was introduced to establish artificial intelligence standards, metrics, and evaluation tools, to support artificial intelligence research, development, and capacity building activities, to promote innovation in the artificial intelligence industry by ensuring companies of all sizes can succeed and thrive, and for other purposes.
AI TRANSPARENCY
In June 2023, Congress introduced a bill that would be enforced by the FTC requiring disclosure of the use of generative AI as an automated process of the AI. In particular, any output must include the disclaimer that the output is the product of generative AI.
US/UK AGREEMENT ON AI SAFETY
On April 1, 2024, the U.S. secretary of state announced an agreement to work with the UK to develop AI safety standards.
IP OWNERSHIP
The copyright office has taken the stance that an AI cannot be an author of a work protected by the copyright act.https://copyright.gov/ai/ai_policy_guidance.pdf
The U.S. Patent office has also taken the position that AI cannot be an inventor under the current patent law. With respect, to hybrid creations, where AI tools are used by human intelligence (HI), the office has offered guidance on how to determine the proper HI inventors. https://www.uspto.gov/subscription-center/2024/uspto-issues-inventorship-guidance-and-examples-ai-assisted-inventions
The Patent office updated its guidance for AI assisted innovation on July 17, 2024. It re-emphasized the need for significant human contribution. https://www.uspto.gov/about-us/news-updates/uspto-issues-ai-subject-matter-eligibility-guidance#:~:text=The%20guidance%20update%2C%20which%20goes,are%20particularly%20relevant%20to%20AI
DEEP FAKES PROPOSED LEGISLATION
HR 6943 No AI FRAUD Act
The No AI FRAUD act seeks to address abuses of artificial intelligence by creating a national right of publicity giving everyone a property right in their own likeness and voice. The right lasts up to 10 years after the person’s death. This is a shorter term than some states that provide similar protection and much shorter than the time period proposed in the Senate NO FAKES Act discussed below.
The Act makes it unlawful to simulate a voice or likeness without authorization. Any person/entity that distributes, transmits or otherwise offers a cloning service or publishes performs, a digital voice replica or digital depiction with knowledge that these were not authorized by the individual holding rights to the voice or likeness at issue are subject to statutory damages of at least $5000 or actual damages, plus any profits attributed to the use. Courts may also award punitive damages and attorney fees.
The Act does provide a fair use defense that appears similar to current copyright fair use. The statute of limitations is 4 years from discovery, and the Act will not preempt state law protections.
Proposed Senate Bill: NO FAKES Act
Senator Coons has offered the NO FAKES Act as another method of addressing AI generated deep fakes. The proposed bill has several similarities to the No AI FRAUD Act including the creation of a property right for individuals that extends beyond their death. NO FAKES, however, seems to offer broader protection recognizing the right in image, likeness and voice. It also provides a term of 70 years after the death of the individual as opposed to the 10 year term in the No AI FRAUD act.
The Act prohibits the unauthorized production of digital replicas as well as the publication, distribution or transmission of those digital replicas. Rather than create a fair use type defense, NO FAKES provides a list of exclusions such as use in news, public affairs, sports broadcasts, documentaries, docudrama, historical or biographical works.
For remedies, the Act offers statutory damages of $5000 per violation or actual damages suffered by the injured party. If the conduct was willful, the court may award punitive damages, and the prevailing party will be entitled to reasonable attorney fees. The statute of limitations is shorter at 3 years compared to the No AI act.
SECURITIES REGULATION
The SEC set out regulations requiring transparency when AI is used by brokers in predictive analytics, and has already enforced these regulations where the AI is allegedly biased towards recommending the broker's managed funds. https://www.sec.gov/news/press-release/2023-140
In February 2024, some members of Congress responded with a bill that would nullify the SEC regulation. https://www.congress.gov/bill/118th-congress/senate-bill/3735/titles?s=1&r=23#:~:text=A%20bill%20to%20prohibit%20the,advisers%2C%20and%20for%20other%20purposes.
Outside the U.S.
The European Union was the first to offer some framework for regulating AI suggesting that the level of regulation should be proportionate to the risk. https://www.europarl.europa.eu/topics/en/article/20230601STO93804/eu-ai-act-first-regulation-on-artificial-intelligence